Multicomponent Assembly of Supramolecular CoordinationPolygons on a Au(111) SurfaceTao Lin,‡ Xue Song Shang,† Pei Nian Liu,*,† and Nian Lin*,‡

†Shanghai Key Laboratory of Functional Materials Chemistry and Institute of Fine Chemicals, East China University of Science andTechnology, Meilong Road 130, Shanghai, China‡Department of Physics, The Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong, China

*S Supporting Information

ABSTRACT: We have studied the self-assembly of a series of multicomponent cyclicsupramolecular polygons on a Au(111) surface using scanning tunneling microscopyand kinetic Monte Carlo simulations. Our results indicate that while closed polygons orcages efficiently self-assemble in three dimensions at appropriate temperatures andconcentrations, chain structures are formed predominantly and the yield of cyclicpolygon structures is very low in two dimensions independent of temperature andconcentration. This substantial shift in the ring−chain equilibrium can be attributed tosubtle competition between kinetic and thermodynamic controls. Simulations suggestthat on a surface, where the translational and rotational freedom of the molecules arerestricted and intramolecular bond flipping is hindered, the low-energy reactionpathways that are essential for the formation of cyclic structures are blocked. In addition,microscopy images suggest that surface defects provide kinetic traps that further reducethe yield of cyclic structures. Our findings reveal the striking differences between the on-surface self-assembly and solution self-assembly of discrete supramolecular systems.

■ INTRODUCTION

Self-assembly of cyclic structures is of high interest infundamental research and applications. Ring−chain competi-tion in polymerization arises when equilibrium exists betweenthe linear and cyclic forms of oligomers and polymers.1−7

Cyclic assemblies predominate below the so-called criticalpolymerization concentration (CPC), while the concentrationof cyclic species remains constant above the CPC, with theexcess monomers producing mainly linear species.1,2,7,8 Suchcompetition deserves attention because it plays important rolesin numerous synthetic processes, such as polymerizationreactions9−12 and polymer assembly via noncovalent inter-actions.1,2,13−15 For example, if the competition is not properlycontrolled, normally by thermodynamics and/or kineticscontrols, the desired products (cyclic or linear) will not beobtained in high yield.Ring−chain competition is also important for supramolecular

coordination self-assembly when the molecules can beassembled into discrete polygons or cages using metal−ligandcoordination.16−20 During the assembly processes, metastableintermediate products frequently form at the local minima onthe reaction free energy hypersurface (FEHS). Depending onthe morphology of the FEHS, self-assembly may be underkinetic or thermodynamic control under the particular assemblyconditions.17 When assembly is under thermodynamic control,intermediates can often be converted into stable low-energytarget products if the reaction is allowed to proceed for arelatively long time or if annealing is performed at elevatedtemperatures.21−23 If, however, the assembly is under kinetic

control, a metastable product can remain kinetically trapped ina deep local minimum.24−31

Recent work has examined supramolecular self-assembly intwo dimensions (2D), i.e., assembly confined to solid−liquid orsolid−vacuum interfaces.32−35 On one hand, 2D supra-molecular self-assembly shows many of the same characteristicsas the corresponding 3D processes. On the other hand,different behavior is expected as the molecules adsorbed on asurface must overcome translational or rotational energybarriers to find their destiny due to surface−moleculeinteractions;35−40 thus, the molecules have fewer degrees offreedom in the 2D assembly processes than in thecorresponding 3D ones. For example, the translational orrotational motions of intermediate supramolecules on a surfaceare greatly reduced and are even frozen at room temperature. Inaddition, torsional motion around a bond is prohibited for mostspecies adsorbed on surfaces; in other words, molecules ormolecular groups cannot flip on a surface. These limitations areexpected to introduce deep local minima on the FEHS of 2Dconfined self-assembly, leading to the creation of thermody-namic minima or kinetic traps.In the present work, we investigated the 2D self-assembly of

multicomponent supramolecular cyclic structures on a Au(111)surface.41−43 We chose ditopic pyridylporphyrin (A and B) andbis(terpyridine) (C and D) as ligands, which are linked by

Received: August 25, 2013Revised: October 2, 2013Published: October 2, 2013

Article

pubs.acs.org/JPCC

© 2013 American Chemical Society 23027 dx.doi.org/10.1021/jp408504b | J. Phys. Chem. C 2013, 117, 23027−23033

terpyridyl-Fe-pyridyl coordination (Chart 1).44 A and C formthree types of cyclic structures: six-component A3C3, eight-

component A4C4, and 12-component A6C6. A and D form twotypes of cyclic structures: four-component A2D2 and six-component A3D3. B and D form the 12-component cyclicstructure B6D6. In all cases, chain structures are formedtogether with the cyclic ones. We explored the ring−chainequilibrium by analyzing the effects of temperature, reactantconcentrations, and component ratios on the yield of cyclicstructures. We also modeled the 2D supramolecular assemblyusing kinetic Monte Carlo (KMC) simulation. Our resultsshow that self-assembly of cyclic structures differs in severalaspects in 2D and 3D; in particular, there exists deeper kinetictraps for the 2D case. We argue that subtle competitionbetween kinetics and thermodynamics in 2D means thatreaction conditions must be controlled more carefully, andsome types of self-assembly may not proceed at all.

■ METHODExperiments were performed in an ultrahigh-vacuum system(Omicron Nanotechnology) with a base pressure below 5 ×10−10 mbar. A single-crystalline Au(111) substrate was cleanedby argon-ion sputtering and annealing to approximately 900 K.The molecules used in this study shown in Chart 1 werethermally evaporated by a molecular beam evaporator anddeposited onto the Au(111) substrate, which was maintained atroom temperature. The evaporation temperatures for these fourmolecules A, B, C, and D are 610, 590, 550, and 590 K,respectively. Scanning tunneling microscopy (STM) measure-ments were performed at 296 K.Kinetic Monte Carlo (KMC) simulations were performed on

a 100 × 100 square or hexagonal lattice using an algorithmpreviously described45 with a periodic boundary conditionimposed. Initially equal amounts of two kinds of molecules

were deposited randomly onto the lattice. Desorption of themolecules was not allowed. Single molecules were allowed tohop to a nearest-neighbor site or rotate clockwise oranticlockwise by 90° on the square lattice or 60° on thehexagonal one. The rates of these events were determined bythe corresponding energy barriers and temperature. Acoordination bond was formed when a ligand encountered acomplementary ligand in a head-to-head configuration. Thebonds can be dissociated at a rate given by the bond strengthand temperature.

■ RESULTS AND DISCUSSIONExperimental Results. Figure 1 shows representative high-

resolution STM images of the six cyclic structures of A2D2,

A3D3, B6D6, A3C3, A4C4, and A6C6 as well as the correspondingstructural models. In these structures, the porphyrin ligands actas corners and the dogbone-shaped bis(terpyridine) ligands actas linkers. Some of the porphyrin molecules appear bright dueto metalation of their macrocyclic core.44−46 The formation ofthe cyclic structures occurred exclusively after mixing Fe withthe two types of ligands, indicating that the structures arestabilized by terpyridyl-Fe-pyridyl coordination.44 Fe atomswere not resolved in the STM topographs, presumably due toelectronic effects.47 A3C3, A4C4, and A6C6 contain the sameligand combinations, but with different coordination anglesbetween two side linker ligands. For example, the A4C4structure exhibits a rhombus shape with corner angles of 65°

Chart 1. The Molecules Used in This Study (Ligands A−D)and the Corresponding Representations of Their CyclicStructures

Figure 1. STM topographs of cyclic structures and the correspondingstructural models of (a) A2D2, (b) A3D3, (c) B6D6, (d) A3C3, (e) A4C4,and (f) A6C6. All STM topographs are 15 × 15 nm2.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp408504b | J. Phys. Chem. C 2013, 117, 23027−2303323028

and 115°, instead of a square shape with a corner angle of 90°.Similar results were obtained with A2D2 and A3D3. Thisnonideal geometry can be attributed to conformationalflexibility of the on-surface coordination bond42,48 andsymmetry mismatch between the atomic lattice of theAu(111) substrate (6-fold) and the ideal cyclic structures.42

Note that in the A + D system A2D2 and A3D3 are incomparable amounts, whereas in the A + C system, A4C4 is themain products while A3C3 and A6C6 are byproducts of lessquantities. In the later sections, we will not analyze these twostructures.Figure 2a−c presents large views of STM images showing the

structures assembled from three ligand combinations (A + C, A

+ D, and B + D) in the presence of Fe. In all three cases, zigzagchains are abundant, whereas cyclic structures (circled in thefigure) are rare. We counted the numbers of different (cyclicand chain) structures over a total area of 10000 nm2. Thequantitative analysis of the structures showed that yield of thecyclic structures, defined as the percentage of the total amountof porphyrin ligands forming cyclic structures, is only 3.3% forA2D2, 2% for A3D3 and A4C4, and 0.8% for B6D6. The yielddeclined with increasing number of constituents in the cyclicstructure (4, 6, 8, or 12). This trend implies that the probabilityof a ligand adopting a correct configuration required forassembling a large cyclic structure (e.g., appropriate molecularorientation) is lower than that for a smaller structure. We variedthe ratio of the two ligands to optimize the self-assembly of thecyclic structures. Figure 3 shows the structures formed by Aand C when mixed in A:C ratios of 2:1 and 1:2. The assembledstructures are primarily small clusters and short oligomers. Onlythe stoichiometric 1:1 ratio resulted in the formation ofextended polymeric structures (cf. Figure 2b). This observationmanifests the importance of stoichiometry in the multi-component self-assembly. We use the samples of stoichiometric

1:1 ratio when analyzing the products of ligand mixtures in therest of this paper.To explore how kinetics controls the self-assembly processes,

we annealed A + D samples at different temperatures fordifferent durations. Annealing at 400 K for 5 min gave lowyields of the cyclic structures similar to those obtained withoutannealing. Extending the 400 K annealing to 2 h increased theyield of A2D2 more than 2-fold to 7.7% and the yield of A3D3 to2.3% (Figure S1). Further annealing with longer time, however,did not enhance the yield, suggesting that the low yield can beascribed to deep kinetic trapping. In order to overcome thekinetic limitation, we increased the annealing temperature to450 K. Contrary to our expectations, only zigzag chains wereobserved (Figure S1). We suggest that the low yield may beascribed to excessive ligand concentration, since cyclicstructures are formed in high yield in 3D only when the ligandconcentration is below the CPC. Therefore, we investigated A+ D self-assembly over a broad range of concentrations. Noneof the ratios led to a significant increase in cyclic yield (data notshown). These findings suggest that while self-assembly ofpolygons can be highly efficient in 3D, their self-assembly isextremely inefficient in 2D. The formation of chain structures isstrongly favored over that of ring structures in 2D. Apparently,this behavior may be attributed to the nonideal coordinationgeometry since the formation of cyclic structures might bethermodynamically less favored than the chain structures whichcontain ideal coordination ones. However, the B6D6 structuresonly contain ideal coordination geometry, but this structurebehaves similarly as the rest. On the other hand, our previousstudy clearly demonstrated that in the assembly of extended 2Dstructures nonideal geometry also occurs; nevertheless, thiseffect does not hinder the self-assembly of the 2D structureswith high yield. Thus, there must be other reasons accountingfor the shifting of the ring−chain balance.

KMC Simulation. We carried out KMC simulations to gaininsight into the mechanism of 2D self-assembly of the cyclicstructures. We simulated four systems: 4-component A2D2, 6-component A3D3, 8-component A4C4, and 12-componentB6D6. The substrate lattice was 6-fold for A2D2 (Figure 4a),A3D3 (Figure 4b), and B6D6 (Figure 4d) but 4-fold for A4C4(Figure 4c). The translational (rotational) freedom of themolecules was constrained by a diffusion (rotational) barrier ofEd = 0.68 eV (Er = 1.00 eV).40,45 When two complementaryligands (yellow and light blue ends) encountered each other ina head-to-head configuration, a bond formed with a binding

Figure 2. Large-view STM images of the self-assembled structures.(a−c) STM topographs (50 × 50 nm2) showing abundant zigzagchains and low-yield cyclic structures in (a) A + D, (b) A + C, and (c)B + D. Representative cyclic structures are circled. (d) The yields ofthe different cyclic structures.

Figure 3. STM topographs (50 × 50 nm2) showing the self-assembledstructures of A + C mixed in different ratios: (a) A:C = 2:1 and (b)A:C = 1:2.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp408504b | J. Phys. Chem. C 2013, 117, 23027−2303323029

energy of Eb = 0.30 eV. Flipping of the molecules (mirroroperation) was not allowed.Figure 5a−d displays representative simulation results

obtained under conditions comparable to the experimental

ones including annealing temperature and duration, molecularconcentration (percentage of substrate surface area covered bythe molecules), and ligand ratio. Similar to the experiments, thesimulations produced mixtures of cyclic structures and zigzagchain structures. The yields of the cyclic structures from thesimulations were 23.6% for A2D2, 4.2% for A3D3, 10.6% forA4C4, and 0% for B6D6 (Figure 4d). Except for B6D6, thesimulated yields of A2D2, A3D3, and A4C4 are higher than theexperimental ones. Several factors may explain this discrepancy.First, when the samples were annealed at high temperature orwith prolonged annealing duration, the Fe atoms that normallyacted as coordination centers to link two complementaryligands may have nucleated into Fe islands or diffused into theAu crystal. The simulations did not take these effects intoaccount. Second, the coordination angles in the experimentalproducts often deviate from ideal geometry, which may reducethe binding strength. Third, A4C4 formed on a 6-fold substratein the experiments, while it formed on a 4-fold substrate in thesimulations. A 4-fold substrate matches the symmetry of theA4C4 structure and is therefore expected to lead to higher yield.Fourth, molecules in the experiments may have becometrapped at surface defects such as step edges, Au(111)herringbone reconstruction, or Fe islands; in contrast, themolecules in the simulations moved freely on the substrate.Despite these differences, the simulations reproduced the keycharacteristics of the experimental results: most products arezigzag chain structures while cyclic structures form in very lowyield. In this way, both experiments and simulation indicate that

self-assembly of cyclic structures is much less favorable in 2Dthan in 3D, so we conclude that the ring−chain equilibrium isshifted toward chain formation. As in our experiments, weperformed simulations to examine the effects of stoichiometryon self-assembly. We simulated self-assembly processes with Aand C present in ratios of 2:1, 1:1, and 1:2 (always at the sametotal molecular concentration), and we compared the yields ofthe cyclic structure A4C4 obtained. The yield for the twononstoichiometric ratios was approximately one-third of thatfor the stoichiometric 1:1 ratio (Figure S2). These results areconsistent with the experimental observations (Figure 3).

Discussion. Self-assembly in 3D occurs through thestepwise formation of various intermediates that leadprogressively to the final target structures. The (almost)unrestricted translational, rotational, and torsional freedom ofindividual molecules or intermediates in 3D provides manylow-energy pathways on the FEHS. As a result, intermediatescan easily find a low-energy pathway to convert into the targetstructures at ambient or slightly elevated temperatures.However, in 2D self-assembly, many of these low-energypathways are blocked since translational, rotational, andtorsional molecular motion is severely restricted. We proposethat this restriction makes the 2D self-assembly of cyclicstructures extremely inefficient. We illustrate this proposal inScheme 1a: I is an intermediate en route to the target cyclic IV;

I can develop not only into IV but also into intermediates II orIII. In 3D, II or III can easily reach the target structure byflipping around a bond and forming an intramolecular bond, aprocess indicated by the red curved arrows in Scheme 1a. On a2D surface, however, flipping around a bond is greatlyhindered, so II or III must convert back into I in order toreach the target structure IV. Since converting II or III into Iinvolves energetically costly bond breaking, II and III lie in adeep kinetic trap on the FEHS. Thus, we propose that theFEHS of 2D self-assembly is highly corrugated as comparedwith the 3D one.To test this proposal, we monitored time-lapse images of

KMC-simulated self-assembly of cyclic structures. Figure 6shows time-lapse images over 66 s of the assembly process ofA4C4. An intermediate similar to II forms initially (red ellipse inFigure 6a). Subsequently, this intermediate develops into achain structure (Figure 6b−f) instead of a cyclic structure. Onemay expect that if intramolecular bond flipping is allowed, theligand at the upper-right corner in the circled structure inFigure 6a would bound with the ligand at the right side,forming a closed cyclic structure. To examine the effects ofintramolecular bond flipping, we introduced flipping events into

Figure 4. The four cyclic structures simulated using KMC. Eachmolecule consists of two ends (yellow or light blue) linked by abackbone (orange or dark blue). The backbone defines the moleculargeometry. A yellow end could bond with a light blue end only if theyencountered each other in head-to-head configuration.

Figure 5. KMC-simulated self-assembly of (a) A2D2 (30%concentration), (b) A3D3 (30%), (c) A4C4 (36%), and (d) B6D6(36%). All assemblies were subjected to annealing at 400 K for 5 min.(e) Yields of the respective cyclic structures.

Scheme 1. (a) Kinetically Controlled Self-Assembly at LowTemperature, in Which II and III Are Kinetically TrappedIntermediates; (b) Thermodynamically Controlled Self-Assembly at High Temperature, in Which ConfigurationalEntropy Prefers Chain Structures

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp408504b | J. Phys. Chem. C 2013, 117, 23027−2303323030

the simulation by reducing the energy barrier of bond flipping.Again we chose the A4C4 system as the representative modelsystem. Figure 7a displays the simulated yield of the cyclic

structure as a function of the energy barrier. One can find theyield of the cyclic structure increases when reducing the barrier,and when the barrier is 0.8 eV, the yield is almost tripled ascompared to the situation without flipping. (Note in thesimulations only L-shaped monomer can flip, and the flippingof dimer or even larger units was not considered due totechnical difficulty.) Because of computational resourcelimitation, we did not perform the simulation of even low-energy barrier. Despite this limitation, the trend revealed inFigure 7a clearly indicates that the yield of the cyclic structuresis substantially high in 3D self-assembly considering that thetorsional rotation barrier of a C−C bond in free space is about0.1 eV.49 Hence, we conclude that hindered bond flipping in2D significantly slows down the self-assembly of the cyclicstructure. We also argue that bond flipping plays a veryimportant role in the formation of cyclic structures in 3D self-assembly.Performing annealing at higher temperatures is a common

strategy to overcome kinetic controls. At higher temperature,trapped intermediates can gain sufficient energy to escape thekinetic traps and explore the FEHS in search of thethermodynamic minimum. Therefore, annealing may convert

intermediates like II and III in Scheme 1a into thethermodynamically stable product IV. To test whether raisingtemperature can enhance the yield of the cyclic structures, wesimulated the self-assembly of A3D3 with annealing in atemperatures range from 350 to 550 K for 5 min (blue curve)and 30 min (red curve) (Figure 7b). When the annealingtemperature is below 400 K, time-lapse images of thesimulation (data not shown) reveal that the molecules do nothave sufficient energy to move effectively on the surface, andonce chain structures form, they do not dissolve. Extending thetime from 5 to 30 min can enhance the yield, confirming thatthe low yield can be attributed to kinetic controls. Thesimulated yield can be raised as high as 60% after 30 minannealing at 425 K. In the experiments, however, prolongedannealing only raised the yield to 10%. The neglecting ofvarious experimental details in the simulations as discussedbefore presumably accounts for this discrepancy. Above 425 K,the yield declines with increasing temperature. At this stage, theself-assembly has reached thermodynamic equilibrium while theformed cyclic structures are frequently dissociated, as can beseen in the time-lapse simulation images. The Jacobson−Stockmayer theory predicts that the equilibrium between cyclicand chain polymers is determined solely by the change inconfigurational entropy associated with cyclization.50 Thus, atsufficiently higher temperatures, the ring−chain equilibrium isexpected to shift toward the chain structures, as illustrated inScheme 1b, because entropy contribution becomes importantat higher temperature. We attribute the decline of the yield atthe high temperature regime to thermodynamic control.The CPC appears to be a critical factor in the ring−chain

competition during cyclic product formation. Therefore, weperformed KMC simulations of the 2D self-assembly of A3D3 atdifferent molecular concentrations (modeled in the simulationas surface area coverage). The simulated annealing was carriedout at 450 K for 30 min to ensure that the reaction would be inthermodynamic equilibrium. At the extreme of very lowconcentrations, the molecules rarely encountered one anotherand therefore rarely formed cyclic structures (Figure 8a). At the

Figure 6. Simulated time-lapse images of KMC-simulated 2D self-assembly of A4C4 with annealing at 375 K. Red ellipses track theevolution of a typical intermediate structure.

Figure 7. (a) KMC-simulated yield of cyclic structures of A4C4 (30%concentration, annealing at 375 K for 30 min) as a function ofintramolecular bond flipping barrier. (b) KMC-simulated yields of the2D self-assembly of cyclic A3D3 (24% concentration) as a function ofannealing temperatures for 5 min (blue) and 30 min (red).

Figure 8. KMC-simulated self-assembly of A and D ligands at differentconcentrations: (a) 1.5%, (b) 18%, (c) 54%. (d) Yields of A3D3 as afunction of concentration. (e) STM image (100 × 100 nm2) showingthe assembly of A and D present at concentration below the CPC.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp408504b | J. Phys. Chem. C 2013, 117, 23027−2303323031

extreme of very high concentrations, the molecules could notmove freely on the substrate for lack of available space, againpreventing them from forming cyclic species (Figure 8c). Atintermediate concentrations, these effects were less severe, andthe yield of cyclic product was relatively high (Figure 8b).Analysis of the yield of the cyclic structures as a function ofconcentration (Figure 8d) shows that the yield rises quickly atlow concentrations, reaching a maximum of 60% yield at 18%coverage, and then declining at high concentrations. Theseresults are consistent with the studies of the effects of CPC onring−chain competition. Experimentally, we did not observe asignificant yield enhancement over a wide range of ligandconcentrations including below the CPC. The STM dataprovide us some clues why this happens: Figure 8e is arepresentative STM image of the self-assembly in which themolecule concentration is below the CPC. One can see that themolecules are trapped at the elbow sites of the Au(111)herringbone reconstruction or step edges. Thus, we argue thatthe surface defects, which effectively trap molecules, hinder thehigh-yield formation of the cyclic structures below the CPC.

■ CONCLUSIONSWe have investigated the self-assembly of a series of cyclicstructures on a Au(111) surface using STM and KMCsimulation. The yield of closed polygon structures in 2D ismuch lower than in 3D. KMC simulations suggest that this lowyield can be attributed to a subtle competition between kineticand thermodynamic control: chain structures predominate atlow temperature due to deep kinetic traps arising from of therestricted molecular motion in 2D, whereas at high temper-ature, thermodynamics (entropy) favors chain structures. Thisdilemma inherently constrains the formation of the cyclicstructures and consequently shifts the ring−chain equilibrium.We propose that to drive the systems to a thermodynamicequilibrium state, one must reduce surface−molecule inter-action, i.e., open the low-energy reaction pathways on theFEHS.

■ ASSOCIATED CONTENT*S Supporting InformationSynthesis sample of A + D under series annealing treatmentsand simulation of A + C with different A:C ratios. This materialis available free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Authors*Tel 852-23587494; e-mail phnlin@ust.hk (N.L.).*Tel 021-64250552; e-mail liupn@ecust.edu.cn (P.N.L.).NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported financially by the Hong KongResearch Grants Council (project no. 602712), the NationalNatural Science Foundation of China (project nos. 21172069and 21190033), the Innovation Program of Shanghai MunicipalEducation Commission (project no. 12ZZ050), and theFundamental Research Funds for the Central Universities.

■ REFERENCES(1) Ten Cate, A. T.; Kooijman, H.; Spek, A. L.; Sijbesma, R. P.;Meijer, E. W. Conformational Control in the Cyclization of Hydrogen-

Bonded Supramolecular Polymers. J. Am. Chem. Soc. 2004, 126, 3801−3808.(2) De Greef, T. F. A.; Ercolani, G.; Ligthart, G. B. W. L.; Meijer, E.W.; Sijbesma, R. P. Influence of Selectivity on the SupramolecularPolymerization of AB-Type Polymers Capable of Both A·A and A·BInteractions. J. Am. Chem. Soc. 2008, 130, 13755−13764.(3) Xu, J.; Fogleman, E. A.; Craig, S. L. Structure and Properties ofDNA-Based Reversible Polymers. Macromolecules 2004, 37, 1863−1870.(4) Ohga, K.; Takashima, Y.; Takahashi, H.; Kawaguchi, Y.;Yamaguchi, H.; Harada, A. Preparation of Supramolecular Polymersfrom a Cyclodextrin Dimer and Ditopic Guest Molecules: Control ofStructure by Linker Flexibility. Macromolecules 2005, 38, 5897−5904.(5) Bastings, M. M. C.; de Greef, T. F. A.; van Dongen, J. L. J.;Merkx, M.; Meijer, E. W. Macrocyclization of Enzyme-BasedSupramolecular Polymers. Chem. Sci. 2010, 1, 79−88.(6) Xiao, T.; Feng, X.; Ye, S.; Guan, Y.; Li, S.-L.; Wang, Q.; Ji, Y.;Zhu, D.; Hu, X.; Lin, C.; Pan, Y.; Wang, L. Highly Controllable Ring−Chain Equilibrium in Quadruply Hydrogen Bonded SupramolecularPolymers. Macromolecules 2012, 45, 9585−9594.(7) Ercolani, G.; Mandolini, L.; Mencarelli, P.; Roelens, S.Macrocyclization under Thermodynamic Control. A TheoreticalStudy and Its Application to the Equilibrium Cyclooligomerizationof β-Propiolactone. J. Am. Chem. Soc. 1993, 115, 3901−3908.(8) Ercolani, G. Physical Basis of Self-Assembly Macrocyclizations. J.Phys. Chem. B 1998, 102, 5699−5703.(9) Gordon, M.; Temple, W. B. Ring-Chain Competition Kinetics inLinear Polymers. Makromol. Chem. 1972, 152, 277−289.(10) Kricheldorf, H. R.; Bohme, S.; Schwarz, G. Macrocycles. 17. TheRole of Cyclization in Kinetically Controlled Polycondensations. 2.Polyamides. Macromolecules 2001, 34, 8879−8885.(11) Kricheldorf, H. R.; Bohme, S.; Schwarz, G.; Kruger, R.-P.;Schulz, G. Macrocycles. 18. The Role of Cyclization in Syntheses ofPoly(ether-sulfone)s. Macromolecules 2001, 34, 8886−8893.(12) Jia, Z.; Monteiro, M. J. Cyclic Polymers: Methods andStrategies. J. Polym. Sci., Part A: Polym. Chem. 2012, 50, 2085−2098.(13) Hofmeier, H.; Hoogenboom, R.; Wouters, M. E. L.; Schubert,U. S. High Molecular Weight Supramolecular Polymers ContainingBoth Terpyridine Metal Complexes and Ureidopyrimidinone Quad-ruple Hydrogen-Bonding Units in the Main Chain. J. Am. Chem. Soc.2005, 127, 2913−2921.(14) Wilson, A. J. Non-covalent Polymer Assembly Using Arrays ofHydrogen-Bonds. Soft Matter 2007, 3, 409−425.(15) Chen, C.-C.; Dormidontova, E. E. Ring-Chain Equilibrium inReversibly Associated Polymer Solutions: Monte Carlo Simulations.Macromolecules 2004, 37, 3905−3917.(16) Chakrabarty, R.; Mukherjee, P. S.; Stang, P. J. SupramolecularCoordination: Self-Assembly of Finite Two- and Three-DimensionalEnsembles. Chem. Rev. 2011, 111, 6810−6918.(17) Cook, T. R.; Zheng, Y.-R.; Stang, P. J. Metal-OrganicFrameworks and Self-Assembled Supramolecular CoordinationComplexes: Comparing and Contrasting the Design, Synthesis andFunctionality of Metal-Organic Materials. Chem. Rev. 2013, 113, 734−777.(18) Fujita, M.; Yazaki, J.; Ogura, K. Preparation of a MacrocyclicPolynuclear Complex, ((en)Pd(4,4′-bpy))4(NO3)8, Which Recognizesan Organic Molecule in Aqueous Media. J. Am. Chem. Soc. 1990, 112,5645−5647.(19) Stang, P. J.; Cao, D. H. Transition Metal Based CationicMolecular Boxes. Self-Assembly of Macrocyclic Platinum(II) andPalladium(II) Tetranuclear Complexes. J. Am. Chem. Soc. 1994, 116,4981−4982.(20) Sun, Q.-F.; Iwasa, J.; Ogawa, D.; Ishido, Y.; Sato, S.; Ozeki, T.;Sei, Y.; Yamaguchi, K.; Fujita, M. Self-Assembled M24L48 Polyhedraand Their Sharp Structural Switch upon Subtle Ligand Variation.Science 2010, 328, 1144−1147.(21) Zheng, Y.-R.; Stang, P. J. Direct and Quantitative Character-ization of Dynamic Ligand Exchange between Coordination-Driven

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp408504b | J. Phys. Chem. C 2013, 117, 23027−2303323032

Self-Assembled Supramolecular Polygons. J. Am. Chem. Soc. 2009, 131,3487−3489.(22) Sato, S.; Ishido, Y.; Fujita, M. Remarkable Stabilization ofM12L24 Spherical Frameworks through the Cooperation of 48 Pd(II)-Pyridine Interactions. J. Am. Chem. Soc. 2009, 131, 6064−6065.(23) Seidel, S. R.; Stang, P. J. High-Symmetry Coordination Cagesvia Self-Assembly. Acc. Chem. Res. 2002, 35, 972−983.(24) Rivetti, C.; Guthold, M.; Bustamante, C. Scanning ForceMicroscopy of DNA Deposited onto Mica: Equilibration versusKinetic Trapping Studied by Statistical Polymer Chain Analysis. J. Mol.Biol. 1996, 264, 919−932.(25) Cui, H.; Chen, Z.; Zhong, S.; Wooley, K. L.; Pochan, D. J. BlockCopolymer Assembly via Kinetic Control. Science 2007, 317, 647−650.(26) Gosal, W. S.; Morten, I. J.; Hewitt, E. W.; Smith, D. A.;Thomson, N. H.; Radford, S. E. Competing Pathways Determine FibrilMorphology in the Self-assembly of β2-Microglobulin into Amyloid. J.Mol. Biol. 2005, 351, 850−864.(27) Hagan, M. F.; Elrad, O. M.; Jack, R. L. Mechanisms of KineticTrapping in Self-Assembly and Phase Transformation. J. Chem. Phys.2011, 135, 104115.(28) Baletto, F.; Doye, J. P. K.; Ferrando, R. Evidence of KineticTrapping in Clusters of C60 Molecules. Phys. Rev. Lett. 2002, 88,075503.(29) Royall, C. P.; Williams, S. R.; Ohtsuka, T.; Tanaka, H. DirectObservation of a Local Structural Mechanism for Dynamic Arrest. Nat.Mater. 2008, 7, 556−561.(30) Korevaar, P. A.; George, S. J.; Markvoort, A. J.; Smulders, M. M.J.; Hilbers, P. A. J.; Schenning, A. P. H. J.; de Greef, T. F. A.; Meijer, E.W. Pathway Complexity in Supramolecular Polymerization. Nature2012, 481, 492−496.(31) Boekhoven, J.; Poolman, J. M.; Maity, C.; Li, F.; van der Mee,L.; Minkenberg, C. B.; Mendes, E.; van Esch, J. H.; Eelkema, R.Catalytic Control over Supramolecular Gel Formation. Nat. Chem.2013, 5, 433−437.(32) De Feyter, S.; de Schryver, F. C. Self-Assembly at the Liquid/Solid Interface: STM Reveals. J. Phys. Chem. B 2005, 109, 4290−4302.(33) Stepanow, S.; Lin, N.; Barth, J. V. Modular Assembly of Low-Dimensional Coordination Architectures on Metal Surfaces. J. Phys.:Condens. Matter 2008, 20, 184002.(34) Lin, N.; Stepanow, S.; Ruben, M.; Barth, J. V. Surface-ConfinedSupramolecular Coordination Chemistry. Top. Curr. Chem. 2009, 287,1−44.(35) Barth, J. V. Molecular Architectonic on Metal Surfaces. Annu.Rev. Phys. Chem. 2007, 58, 375−407.(36) Schunack, M.; Linderoth, T. R.; Rosei, F.; Lægsgaard, E.;Stensgaard, I.; Besenbacher, F. Long Jumps in the Surface Diffusion ofLarge Molecules. Phys. Rev. Lett. 2002, 88, 156102.(37) Otero, R.; Hummelink, F.; Sato, F.; Legoas, S. B.; Thostrup, P.;Lægsgaard, E.; Stensgaard, I.; Galvao, D. S.; Besenbacher, F. Lock-and-Key Effect in the Surface Diffusion of Large Organic Molecules Probedby STM. Nat. Mater. 2004, 3, 779−782.(38) Lauhon, L. J.; Ho, W. Single Molecule Thermal Rotation andDiffusion: Acetylene on Cu(001). J. Chem. Phys. 1999, 111, 5633−5636.(39) Buchner, F.; Xiao, J.; Zillner, E.; Chen, M.; Rockert, M.; Ditze,S.; Stark, M.; Steinruck, H.-P.; Gottfried, J. M.; Marbach, H. Diffusion,Rotation, and Surface Chemical Bond of Individual 2H-Tetraphe-nylporphyrin Molecules on Cu(111). J. Phys. Chem. C 2011, 115,24172−24177.(40) Li, Y.; Lin, N. Combined Scanning Tunneling Microscopy andKinetic Monte Carlo Study on Kinetics of Cu-Coordinated Pyridyl-Porphyrin Supramolecular Self-Assembly on a Au(111) Surface. Phys.Rev. B 2011, 84, 125418.(41) Yokoyama, T.; Yokoyama, S.; Kamikado, T.; Okuno, Y.;Mashiko, S. Selective Assembly on a Surface of SupramolecularAggregates with Controlled Size and Shape. Nature 2001, 413, 619−621.(42) Heim, D.; Seufert, K.; Auwarter, W.; Aurisicchio, C.; Fabbro, C.;Bonifazi, D.; Barth, J. V. Surface-Assisted Assembly of Discrete

Porphyrin-Based Cyclic Supramolecules. Nano Lett. 2010, 10, 122−128.(43) Wang, W.; Wang, S.; Hong, Y.; Tang, B. Z.; Lin, N. SelectiveSupramolecular Assembly of Multifunctional Ligands on a Cu(111)Surface: Metallacycles, Propeller Trimers and Linear Chains. Chem.Commun. 2011, 47, 10073−10075.(44) Shi, Z.; Lin, N. Structural and Chemical Control in Assembly ofMulticomponent Metal-Organic Coordination Networks on a Surface.J. Am. Chem. Soc. 2010, 132, 10756−10761.(45) Gottfried, J. M.; Marbach, H. Surface-Confined CoordinationChemistry with Porphyrins and Phthalocyanines: Aspects ofFormation, Electronic Structure, and Reactivity. Z. Phys. Chem.2009, 223, 53−74.(46) Li, Y.; Xiao, J.; Shubina, T. E.; Chen, M.; Shi, Z.; Schmid, M.;Steinruck, H.-P.; Gottfried, J. M.; Lin, N. Coordination and MetalationBifunctionality of Cu with 5,10,15,20-Tetra(4-pyridyl)porphyrinToward a Mixed-Valence Two-Dimensional Coordination Network.J. Am. Chem. Soc. 2012, 134, 6401−6408.(47) Classen, T.; Fratesi, G.; Costantini, G.; Fabris, S.; Stadler, F. L.;Kim, C.; de Gironcoli, S.; Baroni, S.; Kern, K. Templated Growth ofMetal−Organic Coordination Chains at Surfaces. Angew. Chem., Int.Ed. 2005, 44, 6142−6145.(48) Heim, D.; Ecija, D.; Seufert, K.; Auwarter, W.; Aurisicchio, C.;Fabbro, C.; Bonifazi, D.; Barth, J. V. Self-Assembly of Flexible One-Dimensional Coordination Polymers on Metal Surfaces. J. Am. Chem.Soc. 2010, 132, 6783−6790.(49) Pauling, L. The Nature of Bond Orbitals and the Origin ofPotential Barriers to Internal Rotation in Molecules. Proc. Natl. Acad.Sci. U. S. A. 1958, 44, 211−216.(50) Jacobson, H.; Stockmayer, W. H. Intramolecular Reaction inPolycondensations. I. The Theory of Linear Systems. J. Chem. Phys.1950, 18, 1600−1606.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp408504b | J. Phys. Chem. C 2013, 117, 23027−2303323033